Imagine if, a few years from now, you’re in a meeting. (Even in science fiction, we spend most of our time in meetings.) Everyone’s phones are on the table; your employee badges hang taut around your necks. You start to interrupt your coworker when all the phones chime at once. Without glancing, you know the Meeting Mediator has called a foul on you: it’s someone else’s turn to speak. While checking your email in a fit of pique, you receive an automated request from HR to introduce your colleagues Kavitha and Sasha over Slack. Evidently, they’re working on the same project but you haven’t met–despite sitting down the hall from each other.
Messages like this were creepy at first, but most of the changes to your office have been for the best. Whoever has been rearranging the furniture at night has made it easier for teams to gather and chat. You’ve met more peers in the last six months than in the first three years of working here, thanks to the rotating coffee machines that replaced the single kitchen for the entire company–a dumb idea inspired by an apocryphal story that the placement of Pixar’s bathrooms was designed to create more human interaction. Amazingly, without you really noticing, your once-burning itch to quit has finally cooled.
If this future comes to pass, it’ll be thanks to the box of sensors slung around your neck masquerading as your ID. These “sociometric badges” already exist, created by a Boston-based company called Humanyze. Using a combination of microphones, infrared sensors, accelerometers, and Bluetooth, they measure wearers’ movements, face-to-face (and badge-to-badge) encounters, speech patterns, vocal intonations, and even posture to measure office statistics, like who’s really talking to whom, for how long, and where.
Armed with this information, clients such as Bank of America and Deloitte are in turn mapping these office behaviors to the metrics that matter: sales, revenues, retention rates. You may have already met your quantified self; now say hello to the quantified org.
Humanyze is hardly alone in bringing sensors to bear on the office, but its pedigree and approach stand out in a crowded field. The badges are the product of nearly a decade of research at the MIT Media Lab into the nearly subliminal signals buried in our speech. They represent a massively counterintuitive bet that what we say to each other is much less important than the tonality, pitch, and body language of how we say it, a proposition borne out over hundreds of published papers and experiments.
The badges are roughly the size and weight of a deck of cards, although each iteration has been smaller than the last. They’re worn at sternum height to guarantee clear sound and sight lines for the sensors. (Your conversations aren’t being recorded, they assure you, but your metadata is.) Data is collected and stored locally until the badges are shipped back to Humanyze, where it is then anonymized and analyzed. Results are delivered twice, once as a private readout to each participant–contractually screened from management’s prying eyes–and again as an aggregated report indicating which behaviors or encounters make a difference to the bottom line.
CEO and co-founder Ben Waber describes his company’s approach as “Moneyball for business,” a catchphrase with special resonance for the analytics-obsessed owner of the NBA’s Sacramento Kings. Vivek Ranadivé made a fortune in Big Data before it was big, as the founder of analytics company TIBCO, and while his team floundered on the court last season, he searched for an edge elsewhere. To that end, last October the Kings sent their sales staff into the stands before and during games while wearing the badges. They discovered two things: the reps who spent the most time and energy in motion (both through the stands and in front of customers) sold twice as much as their peers; and the less they talked, the more they sold. Taking these lessons to heart, the Kings tripled their in-game sales last season compared to the previous year. The team has since abandoned cold-calling for face-to-face interactions, quadrupled its sales staff, and started a mentoring program to teach new recruits.
True to the spirit of Moneyball, Humanyze specializes in debunking conventional wisdom around performance, although typically in an office rather than an arena. Its favorite example comes from one of Bank of America’s call centers, which suffered form the usual problems of burnout and higher turnover. A stint wearing badges revealed that the most productive workers frequently shared tips and frustrations with their colleagues. So the company recommended ditching individually staggered breaks in favor of 15 minutes of shared downtime. This supposedly less efficient arrangement–no one is manning the phones–led to shorter calls and lower stress while increasing productivity by more than 10%. “If you can use data to figure out things that are pinpoint small and easy to implement,” says Waber, “they can have order of magnitude effects.”
Two of his favorite tools are cafeteria tables and coffee machines. In one case, simply increasing the size of table from four people to 12 and instituting company-wide lunch hours led to individual productivity increases as high as 25%, thanks to better communication within teams and larger social networks. In another engagement, Humanyze helped Cubist Pharmaceuticals (since acquired by Merck) increase sales by 20 percent, or $200 million. Badge data revealed when Cubist’s sales force increased their interactions with coworkers on other teams by 10%, their sales also grew by 10%. To increase mingling among teams, the company replaced many small coffee stations with several larger ones, imperceptibly seeding the encounters it hoped to see.
What separates Humanyze’s work from just-so stories about office layouts or policies that make workers more productive is linking behaviors with the underlying metrics. Sometimes more coffee machines are necessary, and sometimes they aren’t; it all depends on your objectives. In one instance of best intentions gone astray, Waber’s team was asked to analyze the new open plan office of an unnamed furniture manufacturer. The company had switched cubicles to unassigned seating in hopes of goosing collaboration across teams. It worked, sort of. Interactions increased 17%, but workers’ movement levels also dropped by an average of 14%, indicating that few were standing up and walking around. With teams now scattered around headquarters, this meant that communication between team members had actually crashed by 45%, taking productivity and revenue along with it.
That study was performed with just 65 badges, which has been typical of Humanyze’s short-term consulting engagements thus far. That’s about to change, however, as the five-year-old company begins large-scale, open-ended deployments. Instead of loaning out a few dozen badges for a few weeks at a time–as Fast Company did this year (more on that here)–it will continuously collect data from thousands of workers at its largest customers. Waber is cagey about their identities beyond the handful of public examples we’ve discussed here, although he will allow that they include a number of U.S. banks, some Japanese companies, and Deloitte’s Canadian operation, which is currently using badge data to inform the layout of its new headquarters.
The implications of a sociometrically quantified organization are profound. It’s a given that the hierarchical org chart-as-we-know-it doesn’t begin to describe who’s really working together, and how effectively. Hidden in the spaces between reporting lines are informal peer networks that typically manifest themselves in office politics. Efforts to identify, recognize, and promote these networks range from communities of practice 15 years ago to the current fascination with holacracy. With sociometric badges, these “invisible colleges” are plain to see–at least for management, who can scramble to repair fraying ties. Waber points to the billions of dollars wasted annually on failed M&A as the technology’s most valuable use case: every unhappy corporate marriage is unhappy down to the cellular level. Spot it early enough, and you might still have time to save it.
Taking full advantage of these capabilities will require the reinvention of traditional human resources and facilities management into something more along the lines of Google’s much-touted “people analytics” group, which subjects workplace experiments to the same sort of rigorous A/B testing Marissa Mayer once applied to the color of the logo. “These companies are so data-driven when it comes to their customers, but when it comes to their own people, they’re incredibly lax on that,” says Waber.
Not for long, and not necessarily for the better. While digital surveillance tools are typically described as Orwellian, in this case the better comparison is Frederick Winslow Taylor, the early 20th-century management consultant whose stopwatch-timed dissections of assembly lines led to the de-skilling and ultimately de-humanization of labor. Applied to the office, Taylorism produced factories for paperwork where the emphasis was on throughput and desks were arranged to both minimize interruptions and internalize the watchful gaze of superiors from their offices ringed around the periphery.
Critics have already coined the phrase “digital Taylorism” to describe the kinds of practices documented by Esther Kaplan in the March issue of Harpers, including an all-consuming focus on quotas and efficiency that compels UPS’ delivery truck drivers to cut corners on safety in order to make their numbers. The New York Times’ recent deep dive into Amazon’s toxic culture of highly paid burn-and-churn illustrates how the quantified org will be brought to bear on knowledge workers as well.
“That’s the fear, and it’s a legitimate one,” Waber says. “Most of these applications aren’t aggregating data in a way to make substantial changes in companies; they’re being used to micro-manage people.” They’re also self-defeating, he believes. Tesco’s employees may re-stock the shelves a little faster when they’re wearing armbands that track their every move, but the corresponding plunge in morale and increased turnover ultimately isn’t worth it. Neither is Amazon’s talent hemorrhage by design in pursuit of a faster competitive metabolism—the gains of which Facebook and Asana co-founder Dustin Moskowitz has criticized as illusory.
Humanyze makes the opposite pitch: now that we have the tools to measure what was previously immeasurable, we can manage for values other than efficiency or productivity. When it comes to workspace design, for example, sociometric badges can reveal which layouts are conducive to the kinds of collaborations that matter to actual performance, rather than simply lurching from one fad to the next. (Full disclosure: Waber and I made exactly this argument in the October 2014 issue of Harvard Business Review.)
When it comes to privacy, Humanyze has tried to take a stand by not only anonymizing and withholding individual wearers’ data from their employers, but also threatening litigation should they violate the terms of service, which prohibit trying to reverse engineer that data to figure out, say, which employee is the one skewing the break data by taking two hour lunches. This will become moot, however, once sociometric badges are commodities and organizations bring their own analysts in-house–at which point they can force employees to sign any policies they want. From there, it’s not difficult to imagine a neo-Taylorist office of continuous performance reviews in which bathroom breaks are subject to rigorous statistical analysis and unproductive office friendships are discouraged with the threat of termination. Which is why Waber is one of the few in his nascent industry openly calling for outside regulation before some drastic breach of trust kills the industry in its cradle.
Another is Alexander “Sandy” Pentland, whose Human Dynamics Group at the MIT Media Lab invented sociometric badges nearly a decade ago. Best known as the “godfather” of wearables, including Google Glass (which evolved from prototypes in his lab), Pentland is credited as a co-founder of Humanyze along with his former Ph.D. students Waber, COO Daniel Olguin Olguin, and chief scientist Taemie Kim. But the idea to capture subliminal signals with sensors actually predates them. Working with another named Nathan Eagle (now the CEO of Jana), Pentland distributed special sensor-packed phones to 100 students and faculty living on MIT’s campus and tracked their movements, relationships, moods, and health for nine months in 2004. By the time they were finished, they could predict daily commutes and detect when two people were discussing politics or getting sick.
Pentland has since spun this research into several companies and two books, including last year’s Social Physics. “Social physics seeks to understand how the flow of ideas and information translates into changes in behavior,” he writes in the book’s introduction. He aims to create a “computational theory of behavior,” one that suspiciously sounds like “psychohistory,” the fictional science of galactic-scale probabilistic prediction-making introduced in Isaac Asimov’s Foundation trilogy. (Pentland’s childhood hero was Foundation’s protagonist, Hari Seldon.)
Doing so will require a lot more data, which is why Pentland has championed what he calls the “New Deal on Data,” a proposal to give Internet users complete control over their personal data, which they could then opt to give to services such as Google for Facebook for customization, withhold completely, or sell. Originally developed at the behest of the World Economic Forum in 2008, Pentland has since worked with with Telecom Italia and Telefónica to pilot the scheme in Trento, Italy. But comprehensive legislation has yet to materialize.
Even if strong legal and ethical protections are put in place, the fear persists that our voices will go unheard if we privilege the unspoken. This is the critique of sociometry leveled by British sociologist William Davies in his recent book The Happiness Industry: that the same companies obsessed with employee satisfaction and engagement would never dream of giving them a real voice in how the company should be run.
Waber’s not so sure. “The thing that dominates our work is that what you say, while important, is not that important,” he says. “Who talks to who and how much they talk–now that is powerful.”
Reporting for this story was made possible with the support of the John S. and James L. Knight Foundation.